PDF(917K)

16 Pages·2008·0.89 MB·English

TRENDS AND DIRECTIONS IN CLIMATE RESEARCH
The ENSO Signal in the Stratosphere
Natalia Calvo,a,b Ricardo Garc´ıa-Herrera,b
and Rolando R. Garciaa
aNationalCenterforAtmosphericResearch,AtmosphericChemistryDivision,
Boulder,Colorado,USA
bUniversidadComplutensedeMadrid,CiudadUniversitariaMadrid,Madrid,Spain
AlthoughtheElNin˜o–SouthernOscillation(ENSO)isatroposphericphenomenon,its
effects are also observed in the stratosphere. Traditionally, the study of ENSO above
the troposphere has been difficult because of the lack of global observations at high
altitudesandalsobecauseofthepresenceofothersourcesofvariabilitywhosesignals
are difficult to disentangle from ENSO effects. Recent work with general circulation
modelsthatisolatetheENSOsignalhavedemonstrateditsupwardpropagationintothe
stratosphere.Herewereviewtheliteratureinthisfieldandshowresultsfromthemost
recent version of the Whole Atmosphere Community Climate Model to illustrate the
propagationandthemechanismswherebythesignalmanifestsitselfinthestratosphere.
TheENSOsignalpropagatesupwardtoabout40kmbymeans oflarge-scaleRossby
waves. The propagation is strongly influenced by the zonal mean zonal winds. Most
of the strong ENSO events tend to peak in the boreal winter and so the ENSO signal
is observed mainly at high latitudes during the Northern Hemisphere winter where
thewindsare westerlyandallowRossbywavepropagation.TheENSOsignalisalso
identifiedatpolarlatitudesintheNorthernHemispherewinterintheformofwarmer
temperatures and weaker winds during a strong El Nin˜o event. This signal shows a
zonallyhomogeneousbehaviorfromtheintensificationofthestratosphericmeridional
circulation(inwhichairrisesinthetropicsandmovestowardthewinterpolewhereit
descends)forcedbyanomalouspropagationanddissipationofRossbywavesatmiddle
latitudesduringstrongENSOevents.
Keywords: stratospheredynamics;ENSOvariability;climatology
Introduction been related to changes in the jet stream, the
strong westerly wind current concentrated at
The ElNin˜o–SouthernOscillation(ENSO) middle latitudes in the upper troposphere1,2;
isacoupledphenomenonoftheoceanandthe tovariationsintheMonsoonsystemoverAsia
atmosphere.Itswarmphase,orElNin˜o,refers and Australia3−5; and to changes in cyclone
to the ocean component of the system and is frequencyoverEurope.6
characterizedbyananomalouswarminginthe In the last few decades, many authors have
easterntropicalPacific,usuallyinthenorthern analyzedtheeffectsofENSO,bothinthetrop-
winter. The term Southern Oscillation refers to ics and extratropics. Typically, these studies
the variation in sea level pressure between the have characterized the ENSO phenomenon
eastern Pacific and the Indian Ocean. ENSO by a pressure index, computed as the pres-
isknowntobeoneofthemainsourcesofvari- sure difference between the eastern part and
ability in the tropical troposphere and it has the western part of the Pacific Ocean, or by
using a sea surface temperature (SST) index
obtained by averaging SST over a selected re-
Addressforcorrespondence:Dr.NataliaCalvo,NationalCenterfor
gion in the tropical eastern Pacific. Most au-
AtmosphericResearch,AtmosphericChemistryDivision,P.O.Box3000
Boulder,CO80307-3000.Voice:[email protected] thors have found a response in the tropical
TrendsandDirectionsinClimateResearch:Ann.N.Y.Acad.Sci.1146:16–31(2008).
doi:10.1196/annals.1446.008 (cid:2)C 2008NewYorkAcademyofSciences.
16
Calvoetal.:TheENSOSignalintheStratosphere 17
Figure 1. Microwave Sounding Unit (MSU T2lt) lower tropospheric temperatures for
January 1998. The contour interval is 1 K. Negative contours are dashed. Shadowed re-
gionsindicatetemperatureanomalieshigherthan1Kandshowthecharacteristichorseshoe
pattern.
troposphere that lags the ENSO index by 3– tral part of the Andes. These anomalies that
6 months, depending on the data set and in- occur far away from the tropics are known as
dex used.3,7–11 The spatial distribution of the “teleconnections.”
ENSO signal has also been widely analyzed. Theoretical and model studies have shown
Duringitswarmphase,orElNin˜o,theENSO that the teleconnections are observed because
signal is characterized by an intense warming convective activity in the tropical region gen-
overthecentralandeasternPacificaswellasin erates Rossby waves (large-scale disturbances
most parts of the Indian Ocean and southeast from the variation of the Coriolis force with
ofAfrica.10,12,13 InthetropicalwesternPacific latitude;see,e.g.,Ref.15),and,thus,anomalies
theanomaliesintemperaturetaketheformof inthedistributionofconvectiveactivitygener-
a “horseshoe pattern” with the two extremes atechangesintheatmosphericwavepattern1,16
pointing to the east. Observations of outgoing thatinturnproduceanomaliesintemperature.
long-waveradiation(OLR;theenergyemitted Horel and Wallace1 showed that the ENSO
by the Earth as infrared radiation) and of the teleconnections in the Northern Hemisphere
temperatures measured by the satellite-borne areobservedonlywhenthewindsarewesterly
MicrowaveSoundingUnit(MSU)radiometers at low latitudes (in the winter half of the year
show that maximum warming occurs to the when the Atlantic and Pacific troughs are dis-
eastoftheprecipitationanomaliesthatareob- placedequatorward).ThePacificNorthAmer-
served around the dateline.14 Figure 1 shows ica Pattern (PNA) is among the most impor-
the anomalous pattern in the tropical lower tant teleconnections related to ENSO in the
troposphereforMSUtemperaturesinJanuary Northern Hemisphere,1,16 which in its posi-
1998 during one of the strongest El Nin˜o of tive phase produces above-average geopoten-
thelastcentury.Thehorseshoepatternishigh- tial heights in the vicinity of Hawaii and over
lightedinthisfigureingraywherethetemper- the intermountain region of North America
atureanomaliesarehigherthan1K. and below-average geopotential height south
The impacts of ENSO on temperature are of the Aleutian Islands and over the south-
alsonoticeableatmiddlelatitudes,althoughthe eastern United States. Its relationship with
thermalresponseofthetroposphereinthesere- ENSO is discussed in numerous papers.1,17–19
gionsisnotasintenseasinthetropics.During IntheSouthernHemisphere,thestudiesshow
awarmElNin˜oepisode,ananomalouswarm- anomaliesrelatedtoENSOinhigh-latitudere-
ing is observed in the northwest part of North gions,suchasAntarctica,andinlowerlatitudes
AmericaandthewesternpartofSouthAmer- in Australia and South America.19–23 In addi-
ica,whileananomalouscoolingisregisteredin tion,ENSOalsohasaninfluenceontwomodes
the southeast of North America and the cen- ofvariabilityintheSouthernHemisphere:the
18 AnnalsoftheNewYorkAcademyofSciences
PacificSouthAmericanpattern(PSA)andthe to represent the equatorial wave response to
AntarcticOscillation(AAO).24−27Somestudies anomalous latent heat release.11,14 The zon-
have also shown that the response of atmo- ally symmetricatmospheric warming thatlags
spherictemperaturetoENSOhasanonlinear thedevelopmentoftheSSTanomaliesisinter-
component,thatis,theresponseisnotsymmet- pretedasadiabaticresponsetochangesinthe
ricalforthewarm(ElNin˜o)andcold(LaNin˜a) surfaceenergybalanceduringENSO.10,14
phasesofthephenomenon.28,29 BesidestheENSOsignalinthetroposphere,
ThebehaviorofENSOappearstohavebeen several recent studies have presented evidence
affected by the interdecadal change detected thatENSOpropagatesupwardandalsohasan
in several climatic variables that occurred at effectinthestratosphere.Thislayer,wherethe
the end of the 1970s. On one hand, this temperatureincreaseswithheightmainlyfrom
change modified the development and evolu- theabsorptionofultravioletradiationbyozone,
tionoftheSSTinthetropicalPacific.29–32After spans the range of altitude from about 15 km
1976/1977, the surface temperature warming in the tropics, or 8–10 km in the extratropics,
developed in the western and central Pacific, toabout50km.Itthusliesbetweenthetropo-
extending later toward the eastern part of the sphere (below) and the mesosphere (above). In
basin,10whilebeforethatdatethepositivetem- thispaper,wepresentanoverviewoftheENSO
perature anomalies were first observed along signalinthemiddleatmosphere,togetherwith
the west coast of South America and moved results from a new general circulation model
westward afterward, as shown in Rasmusson that illustrate the propagation of the ENSO
andCarpenter’s4analysisofthe1951–1972pe- signal and the mechanisms by which the sig-
riod. Furthermore, changes in the large-scale nalreachesthestratosphere.Section2reviews
circulation patterns, such as an intensification the literature on the stratospheric ENSO sig-
ofthePNApatternduringstrongENSOevents, nal;Section3introducesthenumericalmodel
have occurred after 1979.33,34 The origin of and methodology used in this paper; and Sec-
thesechangesisnotwellunderstood,although tion 4 presents the main results. Conclusions
some works point to a major influence of the aresummarizedinSection5.
annualcycleduringtheENSOeventsafterthe
1970s,34 while others suggest a modulation of
The ENSO Signal in the
the tropical teleconnections from the Pacific
Stratosphere
Decadal Oscillation related to the SSTs in the
NorthPacific.35 The first works that studied the influence of
ThetemporalevolutionoftheENSOsignal ENSO in the stratosphere, by means of obser-
hasalsobeenwidelystudied.KellyandJones36 vations,werethoseofWallaceandChang,van
identifiedtwopatternsofvariabilityintheSST Loon and Labitzke, Hamilton, Baldwin and
related to ENSO, with a lag of 3 months be- O’Sullivan, and Kodera et al.37−41 However,
tween them. In tropospheric temperature, Yu- the results of these studies are not uniformly
laevaandWallace14alsoidentifiedtwodifferent consistent. Some of the studies show a rela-
patterns lagged by 5 months. More recently, tionship between ENSO and the large-scale
Calvo et al.11 obtained three different spatial cycloniccirculationobservedoverthepolarre-
patterns in tropospheric temperatures related gionintheNorthernHemispherestratosphere,
to ENSO and studied their temporal evolu- usually known as the polar vortex. During a
tion.ENSO-relatedtemperatureanomaliesoc- strong El Nin˜o event, some authors found a
curinitiallyastwodifferentwavepatterns;the weakerpolarvortextogetherwithanintensifi-
firstoneprecedesthefulldevelopmentoftropi- cationofthehigh-pressurecenterlocatedover
calSSTanomaliesandthesecondonedevelops theAleutianregion.Otherauthorsdonotfind
simultaneously with it. These patterns appear anystatisticallysignificantrelationshipbetween
Calvoetal.:TheENSOSignalintheStratosphere 19
ENSO and the stratosphere and they point and the mean flow and, ultimately, the state
out the difficulty of isolating the ENSO signal of the stratosphere in the polar region.38,44,45
fromothersourcesofvariabilitythataffectthe The short records available make it difficult
NorthernHemispherepolarstratosphereinthe to stratify the observations according to QBO
short observational records available (between phases versus ENSO phases and, thus, to an-
20and30years). alyze both phenomena independently. Other
The effects of ENSO on temperature were additional sources of variability that might in-
studied in particular by Reid et al.,42 who terferewiththedetectionoftheENSOsignalin
used several radiosonde stations in the Pacific the stratosphere are volcanic eruptions, which
Ocean, andYulaeva andWallace14 andCalvo inject aerosols into the lower stratosphere and
et al.,11 who used MSU satellite temperatures can change the stratospheric composition and
over the tropical region. They all showed an global circulation46; climate change related to
ENSOsignalinthetropicallowerstratosphere the greenhouse gases, which affects the radia-
of opposite sign to that in the troposphere. tivebalancenearthetropopauseandinfluences
Someauthorssuggestedthatanintensification the local thermal structure and circulation47;
of the meridional circulation in the tropical andsolarvariability,whichcangeneratechem-
troposphere, known as the Hadley cell, with icalanddynamicalchangesinthestratosphere
air rising in the tropics and descending in the andmesosphere.48,49
subtropics, might explain the opposite behav- An additional problem in analyzing the
ior in the troposphere and stratosphere.42,43 ENSO phenomenon in the stratosphere arises
However, Calvo et al.11 explained the tropical fromthelackofglobalobservations.Thecoarse
stratosphericcoolingovertheeasternPacificas horizontal resolution in observations and the
the upper air manifestation of internal equa- fewverticallevelstypicallyavailablehavemade
torial waves forced by anomalous convection thisanalysisdifficult.Thisiswhyinthelastfew
in the troposphere from changes in SSTs. As yearsgeneralcirculationmodels(GCMs),with
wewillshowlater,bothexplanationswererea- aglobalandhomogeneousgridandtheability
sonable as they refer to different behaviors of to isolate different sources of variability, have
theENSOmanifestationintheatmosphere:its becomeoneofthemoreimportanttoolsavail-
zonal-meanbehaviorandthewave-likeENSO able to investigate the ENSO phenomenon in
signal,respectively. the middle atmosphere (the part of the atmo-
Several problems have been found in the spherethatincludesthestratosphereandspans
analysisoftheENSOsignalinthestratosphere. uptoabout100km).
First, ENSO is not one of the main sources of Some of the early modeling studies simply
variability in this atmospheric layer as it is in compared model integrations with and with-
the troposphere, and this makes it more diffi- outimposedSSTanomaliesinthetropicalPa-
culttoidentifyitssignalamongthosegenerated cific. They used models that included only a
by other phenomena. Among the latter is the fewlevelsinthelowerstratosphere.49–55 Even-
Quasi Biennial Oscillation (QBO), an oscilla- tually, other works dealt with the influence of
tion of the stratospheric winds in the tropical SSTvariabilityonthepolarstratosphereinthe
region wherein the equatorial winds oscillate NorthernHemisphere,usingsimplifiedsimula-
from westerlies to easterlies with a period of tionswithmorecompletemodelsthatincluded
about 27 months. The QBO is known to shift a more ambitious treatment of the middle
theregionwherelarge-scaleRossbywavesdis- atmosphere. Hamilton56 reproduced ENSO
sipate in the stratosphere either poleward or variabilityinthestratospherewiththeSKYHI
equatorward,dependingonwhetherthewinds model from the Geophysical Fluid Dynam-
in the tropics are easterly or westerly, and this ics Laboratory in Princeton, New Jersey, using
can affect the interaction between the waves idealized SST perturbations typical of ENSO
20 AnnalsoftheNewYorkAcademyofSciences
eventsduringwintermonths.Theresultswere ofvariability.Thesestudiesdidfindchangesin
inverygoodagreementwiththeobservational the stratospheric circulation related to ENSO
worksdiscussedbeforeandshowedthatENSO and showed the wave-like ENSO signal in the
eventsareassociatedwithstationarylarge-scale stratosphereandthewayitpropagatesupward
Rossbywaveperturbationsofzonalwavenum- fromthetroposphere.Inaddition,theyalsore-
bers1and2.Onemanifestationofthesewave vealed a new aspect not documented before:
anomalies is the intensification of the Aleu- ENSO has an effect on stratospheric zonal
tian High in the middle stratosphere. Lahoz57 mean temperature related to changes in the
studied the possible influence of the SST vari- stratosphericmeanmeridionalcirculation.
ability on the trends in temperature using a Sassietal.59 showedanomaliesthathavethe
tropospheric–stratospheric version of the UK structure of planetary Rossby waves propagat-
Meteorological Office Unified Model. An en- ing up to the mesosphere and studied the mo-
sembleofninesimulationsforwinterconditions mentum deposition by resolved and parame-
intheNorthernHemispherefromthe1980sto terized waves during strong ENSO events in
the1990sshowedthattheUKMetOfficeUni- theNorthernHemisphere.Manzinietal.60 ob-
fied Model reproduced at least part of the ob- tainedasignificantresponseinthezonalmean
servedtemperaturetrend(coolingintheNorth- flow that had not been shown previously, with
ern Hemisphere lower stratosphere) with SST a weaker polar vortex during warm ENSO
variationsbutdidnotfindconclusiveresultsin eventsbutnosignificantsignalfromcoldevents
termsoftheresponsetoENSO. (La Nin˜a). In addition, an enhancement of
In the last few years, several multidecadal thedownwardpropagationofwave–mean-flow
integrations have been run with GCMs. Usu- interaction in the middle atmosphere clearly
ally several realizations of each model are ob- appears in this model. The work of Garcia-
tained and the average of those realizations Herreraetal.61showsgoodagreementbetween
is computed to produce an ensemble average, modelsandreanalysis,whichhighlightsthero-
or ensemble mean. Braesicke and Pyle58 used bustness of the results. While most previous
the UK Met Office Unified Model to com- works focused on the Northern Hemisphere,
paredifferentensembleswithandwithoutSST this work also analyzes the ENSO signal
variabilityandobtainedaweakerpolarvortex in the tropical and Southern Hemisphere
withSSTvariabilitythatdoesnotappeartobe stratospherecomparingdifferentlatitudes.The
related to extreme ENSO events. Some other direct influence of ENSO on the stratospheric
worksspecificallyfocusingonENSOeffectsare circulation through changes in the upward
those of Sassi et al.,59 with the Whole Atmo- propagationofRossbywavesandtherelation-
sphere Community Climate Model, version 1 ship between these two is explicitly shown in
(WACCM1),developedattheNationalCenter this paper for the first time by means of wave
for Atmospheric Research (NCAR); Manzini activityandcirculationvelocitiesanomalies.
et al.,60 using the Middle Atmosphere Euro-
pean Center Hamburg Model MAECHAM5;
and Garcia-Herrera et al.,61 who compared in Model and Methodology
detail results from these two models and the
European Center for Medium-range Weather The WACCM3 is a GCM with interactive
Forecasts (ECMWF) reanalysis of atmospheric chemistry based on the US National Center
observations, ERA-40. None of the models forAtmosphericResearch’sCommunityAtmo-
used in these studies produced a QBO. Al- sphereModel(CAM3).Itincludesmostofthe
though this is a shortcoming of the models, it physical and chemical processes that are im-
doeshavetheadvantagethatitallowsanalysis portantfordescribingthedynamicsandchem-
of model results without this additional source istry of the atmosphere above the troposphere
Calvoetal.:TheENSOSignalintheStratosphere 21
anduptoabout140km.Ithas66verticallev- TABLE 1. Central Month and Its Corresponding
els from surface up to about 140 km, and its N3.4 Value for the Warmest and Coldest ENSO
EventsConsideredintheStudy
vertical coordinate is isobaric above 100 hPa
(approximately 16 km) but hybrid below that WarmENSOEvents ColdENSOEvents
level. The vertical resolution is variable, rang-
January1983 2.74 December1984 −1.21
ing from 1.1 km in the troposphere, above the
January1992 1.82 November1988 −1.91
planetary boundary layer, to about 3.5 km in December1994 1.32 December1998 −1.51
theupperatmosphere. November1997 2.76
The model is used to simulate the period
1950–2004.Asboundaryconditions,SSTsare
prescribed from the global Met Office Hadley for the strongest El Nin˜o and La Nin˜a events
Center,Exeter,U.K.,datasetpriorto1981and were constructed and then the differences (El
fromtheSmith/Reynoldsdatasetafter1981.62 Nin˜ominusLaNin˜aaverages)wereanalyzed.
ThereisnoQBOinthemodel,eitherinternally As noted in the Introduction, different in-
generated or externally prescribed. As noted dices can be used to characterize the ENSO
previously, this may be considered an advan- phenomenon.WehavechosentheNin˜o3.4in-
tage for our purposes as it allows study of the dex(N3.4),whichistheSSTaveragedoverthe
ENSOsignalinisolationfromthisotherimpor- region between long 120◦W and 170◦W and
tantsourceofstratosphericvariability.Chemi- lat 5◦S to 5◦N. Warm and cold ENSO events
caleffectsofvolcanicaerosolsareincludedbut havebeenchosenwheneverthisindexexceeds
not their radiative effects. The 11-year solar 1.2standarddeviations(Table1),asinGarcia-
cycleirradiancevariabilityisparameterizedin Herreraetal.61AlltheENSOeventspeakinlate
termsoftheobservedf10.7radioflux.Detailed fall–earlywinterexceptthestrongwarmENSO
information about the model, processes, and eventinAugust1988,whichhasnotbeencon-
parameterizationscanbefoundinbothGarcia sidered in this study to avoid misleading inter-
etal.63 andRichteretal.64 pretationofthemechanismsinvolvedsincethe
The main differences between WACCM3 stratospheric dynamics is strongly modulated
and WACCM1 (which has also been used to by the seasonal cycle. In the composite analy-
study ENSO effects in the stratosphere, as sis, month 0 indicates the month in which all
notedabove)includetheuseofLin’sfinitevol- the ENSO events reach their maximum N3.4
umeformulation65foradvectioninWACCM3 indexvalue.
as opposed to the semi-Lagrangian method The significance of the anomalies with re-
usedinWACCM1.Thefinitevolumemethod spect to internal natural variability has been
isamass-conservingapproachtothesolutionof computedbyaMonteCarlotestfollowingthe
thegoverningequationsinthemodelandises- methodology of Garcia-Herrera et al.61 In the
peciallyappropriateformodelingtheadvection figures, the shadowed areas denote anomalies
of chemical species. In addition, WACCM3 significantatthe95%confidencelevel.
includes interactive chemistry and solar cy-
cle variability that were not incorporated in
WACCM1.
Results
The results presented here are based upon
threerealizationsofWACCM3runathorizon-
Wave-like ENSO Signal
talresolutionof4◦ ×5◦ (latitude×longitude).
Theensembleaverageofthethreerealizations Figure2displaysthecompositedifference(El
wascomputedfortheperiod1979–2000inor- Nin˜o minus La Nin˜a) for monthly mean tem-
dertobettercomparewithpreviousresultsthat perature anomalies at lat 40oN from month 0
used the same period of analysis. Composites to month 5. Shadowed regions denote the
22 AnnalsoftheNewYorkAcademyofSciences
month=0 month=1
100 1 100 -1 1 -1
-1
80 -2 -1 80 -2 -11
60 -3 -3 1 60 1 -1
-3
40 1 40
-4
-3 1
20 -1 ----1221 20 -2-1
1 1 -1
0 0 -2 1
0 90 180 270 360 0 90 180 270 360
month=2 month=3
100 100
-1 -2 1
80 ---132 1 -11 -2 80 ---3--197---48-1652 -6-7-8 -1
60 --121 60 --4--3--5216
-2 1
40 -3 40
1
1
20 -1 20 1
0 --13-2 1 0 -1-21 -1 -2
0 90 180 270 360 0 90 180 270 360
month=4 month=5
100 -2 -2 100 -1
-3-1 -2
80 -1 1 80 1-2
-4-1--32 1 1 -1
60 --4--321 60
40 1 40 1
1
20 1 1 1 20 1
0 -1 -2 -1 -1 0 -2 -1 -1
0 90 180 270 360 0 90 180 270 360
Figure 2. Composite differences (El Nin˜o minus La Nin˜a) of the temperature anomalies
atlat40oNfromtheWACCM3ensemblesimulationformonths0to5afteranElNin˜oevent.
Solid (dashed lines) denote positive (negative) anomalies. Light (dark) shadowed regions
indicate positive (negative) statistically significant anomalies at the 95% confidence level.
Contoursaredrawnevery1K(zerolinehasnotbeendisplayed).
significant anomalies according to the Monte height,indicatingupwardpropagation.Rossby
CarlotestmentionedinSection3.Thetemper- wavesaregeneratedinthetropospheremainly
atureanomaliesshowtheupwardpropagation bylarge-scaleorography,large-scaleconvective
of the ENSO signal from the Pacific Ocean heating, and land–sea contrasts. From there,
toward the middle atmosphere by means of they propagate into the stratosphere primarily
large-scaleRossbywaves,withnegativeanoma- duringwinterbecauseverticalpropagationre-
lies in the troposphere and positive anomalies quires that the zonal wind be westerly (from
above the tropopause that tilt westward with west to east) and lower than a certain critical
Calvoetal.:TheENSOSignalintheStratosphere 23
value. This threshold depends on the spatial Zonal Mean ENSO Signal
scale of the waves such that only ultralong
In addition to the wave-like signal, ENSO
Rossby waves (zonal wave numbers 1 to 3) are
also generates anomalies in zonal mean fields.
able to propagate into the stratosphere. These
Figure 4 shows the composite differences (El
conditions are known as the Charney–Drazin
criterion.66 Significantanomaliesareobserved Nin˜ominusLaNin˜a)forzonalmeantempera-
ture.WACCM3showsasignificantwarmerpo-
inWACCM3duringthefirst5monthsafterthe
larstratosphereduringastrongElNin˜oevent,
maximumofN3.4indexuptoabout40kmal-
in good agreement with previous works (e.g.,
titude,inverygoodagreementwiththeresults
Refs. 59–61), up to 3 months after the maxi-
from the WACCM1 and MAECHAM5 mod-
els shown in Garcia-Herrera et al.61 but with mum of the N3.4 index, accompanied by an
anomalouscoolinginthetropicalstratosphere.
slightlysmalleranomaliesreachinghigheralti-
The combination of warm and cold anoma-
tudes than for the ERA-40 observational data
set(seeFigure2inGarcia-Herreraetal.61). lies forms a dipole structure that propagates
downward as time progresses, in good agree-
In the middle latitudes of the Southern
ment with results shown in MAECHAM5 by
Hemisphere, the ENSO propagation is not as
Manzinietal.60 Thelargestsignificantanoma-
effectiveasintheNorthernHemisphere(figure
liesinWACCM3,bothinthetropicalandpolar
notshown),mainlybecauseofthetimingofthe
regions, are observed in month 3, with values
ENSOmaximawithrespecttotheseasonalcy-
up to 7 K at high latitudes. In the tropical re-
cle. ENSO events tend to occur in the boreal
gion, the anomalous warming observed in the
winter,whenstratosphericwindsarewesterlyin
troposphere corresponds to the zonally sym-
theNorthernHemisphereandallowtheverti-
metric anomalies that are observed to develop
calpropagationofRossbywaves,inaccordance
during the mature phase of a warm ENSO
with the Charney–Drazin criterion. However,
event.11,14Weakerzonalmeanzonalwindsare
intheSouthernHemispherethewindsareeast-
also observed (figure not shown) at high lati-
erly in the boreal winter and this prevents up-
tudesinthewinterhemisphereduringstrongEl
ward wave propagation. Significant anomalies
Nin˜oeventsaccompaniedbythewarmanoma-
areobservedinthiscaseonlyinthetroposphere
lies,asexpectedfromgeostrophicbalance.
andlowerstratospherebelow20km.
At tropical latitudes, a wave-like ENSO sig-
nal confined to the troposphere and lower
Wave–Mean-flow Interaction
stratosphere that shows composite differences
(El Nin˜o minus La Nin˜a) for monthly mean Thegenerallyacceptedmechanismthatex-
temperatureanomaliesatlat2oNisclearlyob- plains the ENSO signal in the polar winter
servedinFigure3.Thewave-likeENSOsignal stratosphere involves the propagation and dis-
hasoppositesignsinthetroposphereandlower sipation of Rossby waves at middle latitudes
stratosphereaspreviousworkshaveshown.11,14 andtheirinteractionwiththezonalmeanflow.
This is from the small vertical group veloci- In order to analyze in depth this mechanism,
ties of tropical Rossby waves and the easterly we have studied the propagation and dissipa-
winds that inhibit the propagation in this re- tion of the Rossby waves in the stratosphere
gion. Beyond the lower stratosphere, the sig- in WACCM3 and their impact on the strato-
nal acquires a zonal symmetric behavior with sphericbranchofthemeanmeridionalcircula-
the largest negative significant anomalies in tion, also known as the Brewer Dobson circu-
month 3, which is known to be related to lation,whichmovesairfromthetropicallower
the mean meridional circulation as will be ex- stratosphere to the polar regions in the winter
plainednext. hemisphere.Theseanomaliesinthemeridional
24 AnnalsoftheNewYorkAcademyofSciences
month=0 month=1
100 1 100
80 80
1
60 --12 1 1 1-1 60 -2-1 -2 1
-1 -1
1 -1 1
40 -1 40
20 1-1 1 -2 -11 20 -1 -2 -11
0 -11 -21 0 1 -1-2 1
0 90 180 270 360 0 90 180 270 360
month=2 month=3
100 100
80 -1 -11 80 -3 -3 ---2311
1
60 60
40 -1-1 -2 -1-1 40 -1-12 -2
-1
20 -1 -3-2 -2-1 -2 -11 20 --12 -3 -1--211
1
0 11 -1 1 0 -1 1
0 90 180 270 360 0 90 180 270 360
month=4 month=5
100 100
-1 -1 -2 -2
80 --32 80 -1
---123 -11
1
60 1 60 1 1
1-1 1 1-1
40 -2-1 40 -1
20 -2-1 ---3-211 20 -1-2 -3 --12
1 1
0 1 1 0 1 1 1
0 90 180 270 360 0 90 180 270 360
Figure 3. AsFigure1atlat2oN.
circulation are known to generate the anoma- tothezonalmeanflow.Figure5showsthecom-
lies in the zonal mean zonal wind and tem- positedifference(ElNin˜ominusLaNin˜a)ofthe
perature in the polar region as we will explain EP flux anomalies (arrows) and its divergence
next. anomalies(contours)frommonth0tomonth5.
The propagation of the Rossby waves can The upward Rossby wave propagation is in-
be visualized using the Eliassen Palm flux (EP tensified (upward anomalies) from month 1 to
flux),whichcanbeconsideredasameasureof month 5 in the Northern Hemisphere. As ex-
the propagation of wave activity.67 The diver- pected,thewinterclimatology[December,Jan-
gence of the EP flux is a direct measure of the uary,February(DJF)]inWACCM3(Fig.6)for
forcing of the mean flow by the wave pertur- theperiod1979–2000showswesterlywindsin
bations15,67 so that negative values of the EP the Northern Hemisphere and easterly winds
flux divergence indicate regions where Rossby intheSouthernHemisphereduringtheboreal
waves are dissipating and transfer momentum winter. Thus, following the Charney–Drazin
Calvoetal.:TheENSOSignalintheStratosphere 25
month=0 month=1
100 1 1 100
-1 -1 -1
80 80
2
-
60 -1 1 60 1 1 1
1
-1
40 40 -1 -1
-1
20 20
1 1 -1 1
0 0 1
-90 -45 0 45 90 -90 -45 0 45 90
month=2 month=3
100 100
80 -2 -1 -1 80 -3 -2
-1
-1
60 1-1 -3-4 60 -1 1
40 -1 -1 --121 40 -1 -3---245-11
-1
-1 -2
20 20
1 1
0 1 1 1 0 11 1
-90 -45 0 45 90 -90 -45 0 45 90
month=4 month=5
100 100
1
80 -2 -1-2 80 -1
-1 1 -21 1 -2 1
-
60 -1 -2 60
40 1 -1 40 -1 -1
1 -1 1
-1 1
20 -2 1 20 -2 -1
1
0 1 11 0 1 1
-90 -45 0 45 90 -90 -45 0 45 90
Figure 4. AsFigure1forzonalmeantemperatureanomalies.
criterion explained above, westerly winds fa- the composite difference of the anomalies of
∗ ∗
vor the upward wave propagation observed in the mean meridional circulation (V , W ) to-
Figure 5 during strong ENSO events in the getherwiththeanomaliesintheEPfluxdiver-
Northern Hemisphere. Together with the in- gence also shown in Figure 5. From month 1
creasedEPflux,negativeEPfluxdivergenceis tomonth5,aclearintensificationofthestrato-
observedatmiddlelatitudesinthestratosphere spheric circulation is observed, corresponding
with the largest values during months 2 and totheanomalouswavedissipationindicatedby
3 indicating strong wave dissipation in those theEPfluxdivergenceanomalies.Thisacceler-
regions. atedcirculationmovesairfromthelowertropi-
These anomalies in the EP flux divergence calstratospheretothepolarstratosphereinthe
are accompanied by an intensification of the Northern Hemisphere. The rising air in the
meanmeridionalcirculation.Figure7displays tropicsgeneratesadiabaticcoolingasobserved